In some wireless communications systems, such systems comprise a network infrastructure and user equipment, which can for example be portable communications devices. Such communications devices typically receive and transmit signals through the same antenna or set of antennas. This means that some form of duplexing scheme is required in order to allow the device to separate the incoming and outgoing signals such that the former is not swamped by the latter. In this respect, Time-Division Duplexing (TDD) and Frequency-Division Duplexing (FDD) are both well-known duplexing schemes.
Availability of radio spectrum in which to operate the communications system is known to be a limiting factor. So-called 4G, or Long Term Evolution (LTE), is the successor to existing 2G and 3G communications systems. Indeed, LTE-compliant networks are already in operation in many countries. For historical reasons, there are 60 LTE operating frequency bands for the LTE standard as defined in the 3G Partnership Project (3GPP) Rel 15 of the LTE standard, of which 35 require FDD operation. More bands are likely to be defined in later releases of the LTE standard as further mobile broadband spectrum is made available by governments in various territories.
In FDD radio operation, there are two separate carriers at different frequencies, one for the uplink transmission and one for the downlink transmission. Isolation between the downlink and the uplink transmissions is usually achieved by transmission/reception filters called diplexing filters (duplexers or diplexers). These filters are typically implemented as two highly selective filters, one centred on a receive frequency band, the other centred on the transmit frequency band to separate the transmit and receive signals, thereby preventing the transmit signal from interfering with the receive signal. Acoustic resonator filters, such as Surface Acoustic Wave (SAW) filters, are typically used to provide the low insertion loss and sharp roll-off required of duplexing filters. Although these are individually small and cheap, a communications device that is to support multiple frequency bands requires one diplexing filter per frequency band to be supported and further Radio Frequency (RF) switching for selection between the frequency bands so that the duplexing filters can share the antenna.
Furthermore, these filters cannot be integrated with a CMOS circuit owing to the high-Q resonators used to build SAW filters and so they must be implemented off-chip. This is not usually problematic for a simple radio transceiver operating on a single frequency band. However, modern radio transceivers are usually multi-band. As mentioned above, the LTE standard currently specifies 35 FDD frequency bands. To support all of the specified frequency bands would require a manufacturer of user equipment to use multiple filters due to the need for one diplex filter per frequency band supported. A bank of discrete duplexers is one known approach, the bank being connected to an antenna, transmitter and receiver via a multi-way RF switch, which selects the appropriate duplexer based upon a required frequency band of operation. Such an approach increases the complexity of the user equipment, as well as increasing the overall size and cost of the multi-band transceiver. This approach can also lead to performance penalties; for example, the introduction of the RF switch can result in power losses as multiple frequency bands are supported.
Many device manufacturers simply circumvent this problem by designing and manufacturing differently configured devices supporting different sets of frequency bands of operation. Manufacturers thus provide a range of devices each of which is operable in different groups of territories with different frequency band combinations. It can therefore be appreciated that obviating the need for the above-described filters would remove a barrier to the manufacture of a “world phone”, the benefits of which would provide economies of scale to the mobile telephony industry, and mitigate an inconvenience for the international traveller.
Therefore, there is a significant market demand for a solution that is able to replace the fixed tuned diplexer with a flexible device that can support multiple, preferably all, frequency bands.
Although it is possible to tune diplexing filters making up a diplexer, such an approach is currently technically impractical because very high Q-factor resonators are required to achieve the desired selectivity and low power loss. Currently, in order to achieve the small filter size required, such resonators are only realisable as acoustic resonators, which have a well-known bi-resonant characteristic that limits their electrical tuning to only a small frequency range.
An alternative duplexing solution is the use of so-called hybrid junctions or hybrid circuits. This is a 4-port network that can separate the forward and reverse wave directions in a transmission line. Hybrid junctions can be made in a number of ways, including using transformers, waveguides (“magic tees”), or microstrip (“directional couplers”). Hybrid junctions can also be made using active circuits, as is the case for modern electronic analogue wireline phones.
The hybrid junction typically comprises a first (transmit) port, a second (antenna) port, a third (receive) port and a fourth (balance) port. In operation of an ideal hybrid junction, all power incident at the transmit port is divided between the antenna port and the balance port. Likewise, all power incident upon the receive port is divided between the antenna port and the balance port. The device is therefore lossless and reciprocal, and has two symmetry planes with similar properties around each.
Broadband hybrids can be made using transformers, and single-transformer circuits, for example as described in “A Multiband RF Antenna Duplexer on CMOS: Design and Performance” (M. Mikhemar, H. Darabi, and A. A. Abidi, IEEE Journal of Solid-State Circuits, vol. 48, pp. 2067-2077, 2013).
A theoretical hybrid junction, when used as a duplexer, has a power amplifier of a transmitter chain coupled to the transmit port thereof and a low-noise amplifier coupled to the receive port. Transmit power applied at the transmit port by the power amplifier is, as described above, divided between the antenna port and the balance port and the low-noise amplifier is isolated, i.e. there is no leakage of a transmit signal into the receiver as long as the reflection coefficients at the antenna port and the balance port are in balance.
In practice, however, use of the hybrid junction as a duplexer suffers from a number of drawbacks. Firstly, the impedance of the antenna, and so by extension the impedance at the antenna port, typically exhibits variation in both the time domain and frequency domain. The impedance of the antenna can vary with time, for example owing to objects moving in the proximity of the antenna, and consequently, it is necessary to adapt dynamically the impedance at the balance port to the impedance at the antenna port to account for these changes. The antenna impedance also typically varies with frequency and so, to obtain balance at the particular frequency of interest, the impedance at the balance port must be adapted accordingly, and a good balance may be difficult to achieve over a sufficiently wide system bandwidth, for example the 20 MHz needed for an LTE channel. Furthermore, increasing the isolation bandwidth, i.e. the bandwidth over which adequate isolation can be achieved, requires increases in the complexity of the variable balancing impedance circuit, which will increase the size and cost of the device.
Secondly, other coupling mechanisms cause leakage of some of the transmit signal from the transmit port to the receive port of the hybrid junction. As such, isolation of the receive port from the transmit port is limited.
A further technical disincentive to use of the hybrid junction as a duplexer is the absorption of power required to achieve the impedance balance. In this respect, hybrid junctions are typically symmetrical, as mentioned above, with an equal 3 dB loss in each branch of the hybrid junction. Thus, in the context of duplexing, half the transmit power is “wasted” and 3 dB is effectively added to the noise figure by virtue of the waste of received power impacting the signal-to-noise ratio (SNR) of the received signal.
Despite the above-mentioned drawbacks associated with use of the hybrid junction as a duplexer, attempts have been made to obviate or at least mitigate the disadvantages. For example, “Optimum Single Antenna Full Duplex Using Hybrid Junctions” (Laughlin, Beach, Morris and Haine, IEEE Journal of Selected Areas In Communications, Vol. 32, No. 9, September 2014, pages 1653 to 1661), considers an arbitrary antenna with an impedance that can vary widely with frequency and with a return loss that is likely to be of the order of 10 dB minimum (as long as there are no de-tuning proximity effects). This is a practical reality for a transceiver circuit that can be built into a wide range of end products and could possibly be connected through an unknown length of transmission line. So-called Electrical Balance (EB) of the hybrid junction is proposed in the above-referenced document.
However, in FDD systems, including LTE, isolation is required concurrently in both the uplink and downlink bands, meaning that EB duplexers must be simultaneously balanced in both of these bands, which may be separated by more than 100 MHz. This requires either a very wide isolation bandwidth in order to cover both bands, or the EB hybrid junction has to possess a “dual notch” balancing characteristic as was demonstrated in “Tunable CMOS Integrated Duplexer with Antenna Impedance Tracking and High Isolation in the Transmit and Receive Bands” (Abdelhalem, Gudem and Larson, IEEE Transactions on Microwave Theory and Techniques, Vol. 62, no. 9, pp. 2092-2104, September 2014.). Both of these techniques can increase the complexity of the tunable balancing impedance circuit, which may require several or more tunable circuit elements to provide adequate balancing.
For any tunable balancing impedance design, a method and/or apparatus which adjusts the tunable circuit in order to balance the EB duplexer in the required frequency band or bands is required. This is commonly referred to as a balancing algorithm.
European patent publication no. EP-A1-2 903 170 discloses an apparatus for duplexing signals that employs a balancing algorithm. The balancing algorithm is based upon characterisation of, and solution to, a linear equation that describes the transmit-to-receive transfer function of the EB hybrid junction duplexer. A pair of simultaneous equations having two unknowns is disclosed, which can be solved algebraically to yield an expression (a balancing equation) relating a balancing reflection coefficient to a pair of transmit-receive gain measurements made in respect of a respective pair of known balancing reflection coefficients. The balancing algorithm is based upon the balancing equation.
However, the balancing equation described above relies upon knowledge of balancing reflection coefficient values. In particular, the balancing coefficient values need to be set accurately in hardware when the corresponding transmit-receive gains are measured. By way of further explanation, the variable impedance that constitutes the balancing impedance is digitally (or otherwise) set via a control input thereof. In this regard, it has been discovered that a variation exists between the desired setting applied to the control input of the variable impedance, which are, for example, the intended impedance values that are known only from the design of the variable impedance, and actual impedance values achieved, which may differ from the intended design values due to, for example, imperfections in the manufacturing process. When employing the balancing algorithm described above, the isolation achieved between transmit and receive ports of the hybrid junction is suboptimal owing to this difference between the desired impedance and the actual impedance achieved when setting the variable impedance using the digital control. There exists, therefore, a requirement for systems that can balance the duplexer using a similar balancing algorithm, but which is not sensitive to manufacturing imperfections.